Irawan, Bayu (2024) Implementasi Algoritma K-nearest Neighbour Dalam Memprediksi Penyakit Serangan Jantung. Other thesis, Universitas Islam Riau.

Text Bayu Irawan.pdf - Submitted Version Restricted to Registered users only Download (5MB) | Request a copy

Abstract

Heart attack is a deadly disease. The large number of cases and high death rate due to this disease causes the need for a system that can predict heart attacks, so that the death rate can be reduced. This research aims to build a system to predict heart attacks. The K-Nearest Neighbor (KNN) algorithm is a classification method that determines the label (class) of a new object based on the majority of proximity distances between categories in K - Neighbors or nearest neighbors. Model testing was carried out using the K-fold cross-validation (KCV) method to analyze model performance. The data used in testing was 745 for training data and 10 for testing data. Model validity testing was carried out using. To test the kfold cross-validation validation method, the data will be divided into 4 fold test scenarios, namely 5, 10, 15 and 20 fold. Each fold produces an accurate calculation of the confusion matrix. The K-nn values that will be used are 3, 5, 7, 9. From the results of the k-fold cross-validation test, the best k-fold and k-nn values are found at k-fold 20 and k-nn 9 with an accuracy of 67.30 %. After testing the training data, then test the testing data which will be compared with the system results. The comparison results from data testing obtained an accuracy of 70%, precision of 71% and recall of 70%. The hope is that this system can provide assistance to medical personnel who will examine patients who want to have themselves examined.

Actions (login required)

View Item